2.1 Methods Following the previous preliminary study, Goromonzi district of Mashonaland East was retained as the setting for the study. Convenience sampling was used to select respondents to ensure that the sample included both women that had been married before, and those married after 18 years. This would allow for comparison between early and “normal” marriages. Participants for the interviews and focus groups were selected with the help of a local contact person and the support of a partner organisation, the Tree of Life. Thirty-two women agreed to participate in semistructured individual interviews and a further thirty-one women participated in two Focus Group Discussions. For the individual interviews, only 25 interviews could be finally used. Seven forms were lost. The theoretical framework drew from both Phenomenology (Groenewald. 2004)6 and Critical Realism (Easton. 2010)7. This allowed triangulation of the individual interviews, which were phenomenological in nature, with some quantitative questions to produce a balanced understanding and interpretation of the “reality” about marriage in Goromonzi District. The instrument for the interviews (see Appendix 1) used both open-ended questions and specific quantitative questions, had some characteristics of storytelling and a questionnaire. The interviews started with a ‘river of life’ exercise. The exercise was previously used by RAU in earlier research on youth and violence, and was found to be helpful in focusing interviewees on key aspects of their lives8. Here the focus was upon key life experience from the interviewees’ perspective, which would hopefully include aspects of marriage. This was followed by a structured interview, focusing on key points around marriage: before, during and after marriage, including quantitative questions related to each phase in the marriage. Some questions only required a binary – YES/NO – response, whilst others were scored along a four-point Likert scale. The last part of the interview consisted of administration of the Self-Reporting Questionnaire (SRQ-8). This is shortened version of the SRQ-20, developed by the World Health Organisation as an instrument for screening for non-psychotic mental disorders (CSVR, 2006; SACST, 2008). The SRQ-20 has been widely used throughout Africa and other developing countries, and the shortened version was developed specifically for Zimbabwe (Patel & Todd, 1996). All scores of 4/5 or higher on the SRQ-8 are indicative of significant psychological disorder (Patel & Todd, 1996), and the instrument was used as another critical realist measure. The measure was used to examine whether the sample had any significant psychological disorder, and whether this was a consequence of marriage or other life events. It was also possible that the instrument might indicate whether early marriage was more related to such disorder than conventional marriage. As pointed out above, in addition to the interviews FGDs were held to gather commonly held views on the phenomena (child marriage) under investigation. In the first FGD there were 16 participants and in the second group 15 participants. Six hypothetical stories were derived from Groenewald, T (2004), A Phenomenological Research Design Illustrated. International Journal of Qualitative Methods 3 (1), 125. 7 Easton, G (2010), Critical realism in case study research. Industrial Marketing Management, 39, 118–128. 8 Osteroom, M.& Pswarayi, L (2014), Being a Born-free. Violence, Youth and Agency in Zimbabwe. RESEARCH REPORT, Vol. 2014, No, 79. INSTITUTE OF DEVELOPMENT STUDIES. 6 9